High-latency tool calls break agent loops. Learn how to deploy your own instance of the wmcp.sh gateway to Cloudflare Workers for global sub-100ms execution.
When your LLM waits seconds for a remote server to respond to a tool call, the user experience degrades. Traditional container deployments introduce cold starts and routing delays. Deploying wmcp.sh on edge infrastructure eliminates this overhead.
Disclaimer: wmcp.sh is an open-source project and is not affiliated with Cloudflare, Vercel, or Anthropic.
# 1. Clone the wmcp.sh repo
git clone https://github.com/New1Direction/webmcp-anything.git
cd webmcp-anything/worker
# 2. Configure wrangler.toml
name = "wmcp-edge"
main = "src/index.ts"
compatibility_date = "2024-05-28"
[placement]
mode = "smart"
# 3. Deploy to your Cloudflare account
npx wrangler deploy| Capability | Local Docker | wmcp.sh on Cloudflare |
|---|---|---|
| Execution Latency | ⚠️ 500ms+ depending on region | ✅ Sub-100ms global edge |
| Scaling | ❌ Manual container orchestration | ✅ Serverless auto-scaling |
| Caching | ❌ Requires external Redis | ✅ Native short TTL cache (~1s) |
| Cold Starts | ⚠️ Noticeable on first request | ✅ Near-zero cold starts via V8 isolates |
Deploying on Cloudflare Workers ensures sub-100ms latency by running your tool execution gateway close to the language model providers and end-users.
No, you can start on the free tier, but high-throughput agent traffic may require scaling.
No. wmcp.sh is an open-source project and is not affiliated with Cloudflare, Anthropic, or OpenAI.
Custom adapter + hosted MCP at mcp.yourbrand.com + verified badge. From $499 one-time setup. (Managed Retainer: $999/mo, Enterprise: $4,999+/mo)